Output passing

Output passing lets one job publish values that downstream jobs can consume during the same trigger chain.

Why it exists

Without output passing, pipelines tend to rely on:

• hard-coded temp paths
• shared global environment variables
• ad hoc files with unclear ownership

Husky gives outputs first-class scope and persistence.

Declare outputs

jobs:
  ingest_events:
    description: "Download events"
    frequency: manual
    command: "./scripts/ingest.sh"
    output:
      file_path: last_line
      exit_code: exit_code

Capture modes

Mode	Meaning
last_line	Final non-empty stdout line
first_line	First stdout line
json_field:<key>	Parse stdout as JSON and extract a field
regex:<pattern>	Capture the first regex group
exit_code	Store the process exit code

Use outputs in downstream jobs

Outputs can be referenced in:

• command
• env

jobs:
  transform_events:
    description: "Transform the file from ingest"
    frequency: after:ingest_events
    command: "./scripts/transform.sh"
    env:
      INPUT_FILE: "{{ outputs.ingest_events.file_path }}"

Example chain

jobs:
  ingest_events:
    description: "Download raw events"
    frequency: manual
    command: "./scripts/ingest.sh"
    output:
      file_path: last_line

  transform_events:
    description: "Transform raw file"
    frequency: after:ingest_events
    command: "./scripts/transform.sh"
    env:
      INPUT_FILE: "{{ outputs.ingest_events.file_path }}"
    output:
      record_count: json_field:count

  generate_report:
    description: "Report on transformed data"
    frequency: after:transform_events
    command: "./scripts/report.sh"
    env:
      RECORD_COUNT: "{{ outputs.transform_events.record_count }}"

Cycle scoping

Outputs are scoped by cycle_id. That means:

• one manual run of a root job creates one cycle
• downstream jobs in that chain read outputs only from that same cycle
• later runs do not reuse earlier cycle values

Failure behavior

If a downstream template references an output that does not exist for the current cycle_id, dispatch fails immediately with a descriptive error.

Inspection

Use:

husky logs <job>
curl http://127.0.0.1:8420/api/runs/<id>/outputs
curl http://127.0.0.1:8420/api/db/run_outputs

The dashboard also shows captured outputs in run detail views.